Summary

Over the last decade dust models have significantly improved the physical representation of dust emission as well as the location of the sources. However, comparisons of dust model results with long-term observations indicates that models are unable to reproduce observed decadal variability of dust. Such discrepancy may be related to vegetation growth and decay in relation to rainy and drought periods.

In this presentation, after briefly describing our coupled climate models with dust emission from the dynamic vegetation model, we will discuss the results of experiments designed to evaluate the amplification of dust emission by vegetation. We will first show that vegetation and landuse control dust decadal variability using a historical simulation. Interestingly, this simulation reveals an hemispheric asynchronous and asymmetric dust loading. Indeed, in the 50s and 60s dust load was particularly low in the Northern Hemisphere while in the Southern Hemisphere dust activity in Australia was intense. Then, in the 70s the reverse was observed. If we know that Sahel drought is most likely beyond the increase of dust emission in West Africa, there is actually no explanation for Australia. Using a set of simulations with our fully coupled models, we will show that Australian dust is controlled by ENSO through soil moisture and vegetation changes. We will prove our case with satellite data.

To go further, we would like to address the following question: What effect dust loading asymmetry between hemispheres has been playing on tropical rainfall, and potentially the Sahel drought? Analysis of an ensemble runs with our climate coupled models reveals a near "perfect" linear relationship between dust loading asymmetry and shift of tropical rainfall over the Atlantic Ocean. These results have also implications for the analysis of dust impact on the displacement of the Intertropical Convergence Zone during the Holocene (Younger Dryas or Last Glacial Maximum), when dust hemispheric asymmetry was considerably higher.